- Email: [email protected]
Semi-rational approaches to engineering enzyme activity: combining the benefits of directed evolution and rational design
Semi-rational approaches to engineering enzyme activity: combining the benefits of directed evolution and rational design Roberto A Chica1,, Nicolas Doucet2, and Joelle N Pelletier1,2 Many research groups successfully rely on whole-gene random mutagenesis and recombination approaches for the directed evolution of enzymes. Recent advances in enzyme engineering have used a combination of these random methods of directed evolution with elements of rational enzyme modification to successfully by-pass certain limitations of both directed evolution and rational design. Semi-rational approaches that target multiple, specific residues to mutate on the basis of prior structural or functional knowledge create ‘smart’ libraries that are more likely to yield positive results. Efficient sampling of mutations likely to affect enzyme function has been conducted both experimentally and, on a much greater scale, computationally, with remarkable improvements in substrate selectivity and specificity and in the de novo design of enzyme activities within scaffolds of known structure. Addresses 1 De´partement de chimie, Universite´ de Montre´al, CP 6128, Succursale Centre-Ville, Montre´al, Que´bec, H3C 3J7, Canada 2 De´partement de biochimie, Universite´ de Montre´al, CP 6128, Succursale Centre-Ville, Montre´al, Que´bec, H3C 3J7, Canada Corresponding author: Pelletier, Joelle N ([email protected]) These authors made an equal contribution to this work.
Current Opinion in Biotechnology 2005, 16:378–384 This review comes from a themed issue on Protein technologies and commercial enzymes Edited by Bernhard Hauer and Brian K Kay Available online 1st July 2005 0958-1669/$ – see front matter # 2005 Elsevier Ltd. All rights reserved. DOI 10.1016/j.copbio.2005.06.004
Introduction Enzymes are powerful catalysts that are able to increase reaction rates by up to 17 orders of magnitude [1,2]. Certain enzymes display perfect control over stereochemistry and regioselectivity, while others display a breadth of specificity; either feature may be attractive for industrial and synthetic applications [3]. However, few enzymes naturally catalyze the reactions that chemists need under conditions that are industrially convenient and economically advantageous. Many efforts are currently devoted to the modification of enzyme activities to meet the needs of today’s chemists. Multiple approaches have been developed to allow the identification of mutant enzymes possessing desirable Current Opinion in Biotechnology 2005, 16:378–384
qualities such as increased activity, modified specificity, selectivity or cofactor binding [4]. The earliest approach was rational design, which was used to modify the specificity of enzymes [5–9]. This approach requires an indepth knowledge of the structural features of the enzyme active site and their contribution to function. The complexity of the structure/function relationship in enzymes has proven to be the factor limiting the general application of rational design. More recently, directed evolution has proven to be a powerful tool for the modification of enzyme activities and has become the most widely used approach. Briefly, directed evolution consists of the low frequency introduction of randomly distributed mutations in a gene of interest, followed by selection of the mutated proteins possessing the desired properties. Recent advances in this area include the application of the immune system ‘hypermutation’ process, which is conducted entirely in live cells [10]. Directed evolution can enable the relatively rapid engineering of enzymes without requiring an in-depth understanding of structure/ function relationships, unlike rational design. The numerous methodologies and successes of directed evolution will not be detailed here as recent reviews have been devoted to the topic [11–15] (also see the article by AA Henry and FE Romesberg in this issue). Because large numbers of mutants must generally be screened to obtain a significant, desired effect on enzyme activity, the main limitation of directed evolution is the necessity of developing a high-throughput screening methodology that allows identification of the desired property under relevant conditions. Not all enzyme activities are readily amenable to developing a high-throughput screening method, nor are all screening methodologies easy to implement at the required scale. A further limitation of directed evolution relates to the random nature of the mutations introduced: in the case where functional information (from point mutations, random mutagenesis or deduction by sequence alignment) or structural information exists, it would be advantageous to exploit this by concentrating mutations where they might be the most effective [16]. Analysis of enzyme modification results indicates that the majority of mutations that beneficially affect certain enzyme properties (enantioselectivity, substrate specificity and new catalytic activities) are located in or near the active site, and more specifically near residues that are implicated in binding or catalysis [15,17,18,19]. Other properties (stability and activity) can be improved by mutations either near or far from the active site, although there are many more mutations far from the active site to test out [19]. www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 379
Table 1 Comparison of approaches for engineering enzyme activity. Rational design
Random mutagenesis
Semi-rational design
High-throughput screening or selection method Structural and/or functional information Sequence space exploration
Not essential Both essential Low
Essential Neither essential Moderate, random
Probability of obtaining synergistic mutations
Moderate
Low
Advantageous but not essential Either is sufficient Experimental: moderate, targeted Computational: vast, targeted High
Consequently, saturation mutagenesis (where all 20 natural amino acids are tested at residues in or near the active site) can be coupled to recombination of these mutations in order to increase the likelihood of beneficially modifying the catalytic activity relative to random mutagenesis approaches. This ‘semi-rational’ approach could be particularly advantageous in instances where no highthroughput screening method is available, based on the argument that a ‘smarter’ library can be built with the same number of mutants than with a random whole-gene mutagenesis approach [15] (Table 1; Figure 1). This review provides an overview of recent articles where the semi-rational design of enzyme activities
has been successfully demonstrated through the use of combinatorial mutagenesis biased toward the active site, based both on experimental and computational approaches.
Modifying enzyme activity using experimental semi-rational approaches The experimental combination of rational and random protein engineering approaches has been successfully applied toward the modification of enzyme activities. The following examples have been grouped according to the specific experimental approaches applied, which are guided by the amount and nature of structural and functional information available at the outset of the study.
Figure 1
Choice of method for enzyme engineering
Functional information may be gained from small library or point mutations
NO
NO
High-throughput screening available?
Structural/functional information available?
NO
YES
High-throughput screening available?
YES
YES WHOLE-GENE RANDOMIZATION
SEMI-RATIONAL DESIGN Rational design
SEMI-RATIONAL DESIGN Whole-gene randomization Current Opinion in Biotechnology
Selection of the preferred experimental approach for enzyme engineering based on the availability of experimental tools and prior knowledge of structure and function. Rational design, semi-rational design or whole-gene randomization each refer to multiple methodologies, as outlined in the text. The enzyme engineering approach that may have the greatest potential for success is in upper case letters, while alternative approaches are in lower case letters. www.sciencedirect.com
Current Opinion in Biotechnology 2005, 16:378–384
380 Protein technologies and commercial enzymes
Targeted randomization of defined residues based on structural knowledge
Structural information, when available, can allow one to target specific residues in direct contact with the substrate or near the active-site cavity for mutagenesis, alone or in combination. Although the choice of the positions to mutate remains rational in most cases, the choice of amino acids to be encoded can be broad; furthermore, the simultaneous randomization at targeted positions may result in synergistic effects [20] that could not have been predicted by mutating positions individually. Santoro and Schultz [21] have used targeted randomization to modify the substrate specificity of a Cre DNA recombinase from bacteriophage P1 that recognizes DNA sequences known as loxP sites. The researchers used structural information of the Cre–DNA complex to target two regions of the Cre recombinase in contact with loxP base pairs suspected to be important for recombination activity. By creating two distinct libraries (C1 and C2; see Figure 2a) of five and six simultaneously randomized residues in contact with variants of loxP sites (library size 108) and by subjecting the E. coli-transformed libraries to a powerful recombination assay of fluorescent reporter proteins sorted by FACS (fluorescence-activated cell sorting), they isolated a mutant that efficiently recombined a new loxP site not recognized by the wild-type enzyme. Moreover, this Cre mutant retained the ability to recognize the native loxP site. The authors favored this targeted library approach because they anticipated that a higher likelihood of success would be observed, while requiring fewer rounds of selective amplification and
mutagenesis compared with random mutagenesis of the whole gene. Similarly, in a successful attempt to investigate the detoxification role of epoxide hydrolases, Rui et al. [22] used a combination of rational design and saturation mutagenesis at targeted active-site positions to expand the substrate range of an epoxide hydrolase from Agrobacterium radiobacter AD1 (EchA) to include chlorinated epoxides. On the basis of a careful investigation of the active site by structural comparison to related enzymes, residues F108, C248, I219 and I111 (in single-letter amino acid code), in the vicinity of the catalytic triad, were separately randomized by saturation mutagenesis. As mutant C248I showed a slight increase in rate of cis1,2-dichloroethylene (cis-DCE) mineralization (2.7-fold increase with respect to the wild type), it was used as a template for successive rounds of saturation mutagenesis targeting residues F108 and I219. The approach generated the triple active-site area mutant F108L-I219LC248L, which displayed a 10-fold enhancement in cis-DCE mineralization relative to wild-type EchA. Several other groups [23–26] have also successfully conducted simultaneous or successive targeted randomization of multiple active-site positions based on structural knowledge, as an efficient means to circumvent limitations inherent either to site-directed mutagenesis or to whole-gene random mutagenesis. Recently, Schultz and colleagues [27] combinatorially randomized two activesite residues of an aminoacyl-tRNA synthetase as a step in the generation of an orthogonal synthetase/tRNA pair
Figure 2
Residues in or near the enzyme active site that were targeted for semi-rational combinatorial mutagenesis. The enzymes are shown as grey ribbon diagrams with space-filling in semi-transparent gray. (a) Cre DNA recombinase from bacteriophage P1 [21] (PDB accession number 1CRX). Residues from libraries C1 (six residues) and C2 (five residues) and the DNA substrate are colored yellow, green and blue, respectively. (b) PBP2X DD-transpeptidase from S. pneumonia [30] (PDB accession number 1QMF). The ten residues targeted for combinatorial site-directed mutagenesis are colored yellow and the three subsequent unforeseen beneficial mutations obtained by epPCR are colored green. Cefuroxime is acylated at the active-site residue Ser337 and is colored blue. (c) TEM-1 b-lactamase from E. coli [43] (PDB accession number 1FQG). The 19 active-site residues targeted for in silico mutagenesis using PDA are colored yellow. The benzylpenicillin antibiotic is acylated at the active-site residue Ser70 and is colored blue. Figure generated with Pymol Release 0.98 (http://www.pymol.org). Current Opinion in Biotechnology 2005, 16:378–384
www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 381
that efficiently and selectively incorporates an unnatural amino acid into proteins. In some cases, the structural information available is not sufficient to rationally select residues to be randomized. Thus, molecular modelling studies have been successfully used to identify residues most likely to be in contact with substrate molecules. Our group used this approach to perform combinatorial targeted mutagenesis of all 16 principal active-site residues of type II R67 dihydrofolate reductase, allowing the selection of new, highly modified active-site environments with activity at least as great as the native enzyme [28]. Similarly, the substrate specificity of Bacillus stearothermophilus SD1 D-hydantoinase was modified by a semi-rational mutagenesis approach targeting residues chosen on the basis of previous modelling studies [29]. Simultaneous random mutagenesis and site-saturation of defined residues
Based on evolutionary protein-fold similarity between two related enzyme families, Peimbert and Segovia [30] introduced a b-lactamase activity into PBP2X DDtranspeptidase from Streptococcus pneumonia through a combination of error-prone polymerase chain reaction (epPCR) and saturation mutagenesis at targeted positions. This approach was used to increase the odds of obtaining beneficial mutations at unpredictable locations in combination with mutations at carefully chosen residues. This approach should increase the return using the same number of mutants as a random whole-gene mutagenesis scheme. Thus, two active-site residues (F450 and W374) were targeted for saturation mutagenesis in combination with simultaneous amino acid replacements at eight positions near the active-site cavity (see Figure 2b). After the oligonucleotide-based mutagenesis, the gene was further amplified with a 0.8% random mutagenesis rate. This library was selected for cefotaxime resistance and mutants with 10-fold increased resistance relative to wild-type PBP2X were obtained. The random mutagenesis resulted in additional mutations at positions 312, 452 and 554, both proximal to and distal from the active site. To assess the impact of certain randomly inserted mutations, saturation mutagenesis was undertaken at positions 312, 336, 450 and 452. Although no further increase in cefotaxime resistance was observed compared with the original mutants, one new mutant showed b-lactamase activity without compromising the original DD-peptidase activity of PBP2X, thus generating a mutant with dual substrate specificity. Random mutagenesis followed by site-saturation of defined residues
The structural or functional information necessary to make rational choices for the residues to mutate is not always available. To circumvent this limitation, several research groups have efficiently undertaken rounds of www.sciencedirect.com
whole-gene randomization to provide ‘leads’ (i.e. residues identified as potentially advantageous when mutated) in conjunction with fine-tuning of the ‘lead mutants’ through site-saturation mutagenesis of the identified residues. Using such a cyclical random/targeted approach, Reetz and collaborators greatly improved the enantioselectivity (E) of Pseudomonas aeruginosa lipases toward a p-nitrophenyl ester in favour of the (2S) enantiomer (reviewed in [31]). Initial improvement from E = 1.1 (wild-type enzyme) to E = 11.3 was achieved after four rounds of epPCR. Using the mutational knowledge obtained by epPCR, they conducted saturation mutagenesis at given mutational ‘hot spots’ of the lipase, obtaining E = 20. Subsequent rounds of epPCR increased E to 25. In addition, they developed a modified version of Stemmer’s combinatorial multiple-cassette mutagenesis method [32] to mutate a defined region of the lipase (residues 160–163) as well as several ‘hot spots’ (residues 155 and 162). This semi-rational method yielded their most highly enantioselective lipase variant to date (E = 51). Other groups have recently demonstrated the power of similar semi-rational approaches toward the evolution of enantioselective enzymes. As a result of the multiple random mutagenesis steps, this approach has also generated many mutations far from the active-site cavity in addition to mutations designed within the active-site cavity [33–35]. Geddie and Matsumura [36] have modified the substrate specificity of Escherichia coli b-glucuronidase through combinatorial site-saturation mutagenesis of several ‘hot spots’, using knowledge previously obtained from DNA shuffling studies. The resulting mutants displayed up to a 70-fold increase in xylosidase activity and were further improved by several steps of whole-gene random mutagenesis using DNA shuffling, epPCR and Staggered Extension Process (StEP) to generate mutants displaying 100-fold improvement in xylosidase activity compared with their ancestor. Similar studies were undertaken to modify the enzymatic specificity and activity toward 7-aminodesacetoxycephalosporanic acid (7-ADCA) of a glutaryl acylase of Pseudomonas SY-77 [37] and to modify and improve fluorescent proteins from Discosoma sp. [38]. In all of these studies, whole-gene randomization was generally undertaken in a first round of mutagenesis, followed by ‘fine-tuning’ of the most interesting mutants through single or combinatorial site-saturation mutagenesis at targeted positions. Afterwards, mutants obtained were further improved using subsequent rounds of random mutagenesis. In the absence of structural or functional information, the ‘Evolutionary Trace’ method [39] can pinpoint residues to mutate. This method entails correlating evolutionary variations within a gene of interest with divergences in the phylogenetic tree of that sequence family. This has Current Opinion in Biotechnology 2005, 16:378–384
382 Protein technologies and commercial enzymes
been shown to reveal the relative functional importance of residues and to identify functional sites [40]. A recent review by Minshull et al. [41] provides a highly comprehensive overview of this and other approaches for identifying residues that may be functionally relevant. These recent examples demonstrate the power of structure-based semi-rational approaches targeting active-site residues in enzyme engineering by enhancing our capacity for rational design, while exploiting the advantageous aspects of random mutagenesis. We can thus cycle between light sampling of randomly distributed mutations and saturation mutagenesis at a limited number of positions likely to affect the property under study, increasing the likelihood of identifying beneficial, cooperative effects with respect to enzyme catalysis. Semi-rational and combinatorial design using computational approaches
Although semi-rational, combinatorial mutagenesis directed at active-site residues limits the number of variants generated relative to available sequence space and theoretically focuses on an area of sequence space where improvements have a greater likelihood of occurring, it can nonetheless generate greater numbers of mutants than can be screened. Indeed, combinatorial randomization of only five residues generates a library of 205 possibilities (3.2 106 mutants), too large a number for manual screening. Thus, to increase the power of semirational and combinatorial modification of enzyme activities, computational methods have been developed based on protein design algorithms. These methods can either perform a virtual screening of a vast library or can be applied to the design of enzyme active sites. The first method is the computational screening of mutant sequences of a virtual library. Computational screening can screen libraries of 1080 variants [42], allowing for a first layer of virtual screening to eliminate mutations inconsistent with the protein fold. This is an emerging area in protein design and has seldom yet been applied specifically to enzyme design. However, its resounding success and great promise merit discussion herein. Hayes et al. [43] developed a strategy for the computational screening of large libraries called Protein Design Automation (PDA), for prediction of the optimal sequence that can adopt a desired fold. PDA was used for prescreening large, virtual libraries of mutants (1023), thus decreasing the sequence space of interest by many orders of magnitude. PDA allows all, or a rationally defined set of residues, to change. The optimal sequence is chosen based on its lowest conformational energy and is used to identify other near-optimal sequences through Monte Carlo simulated annealing. The mutations that occur most frequently define the library to be experimentally screened. Using PDA, Hayes and colleagues pinpointed 19 residues of interest in TEM-1 b-lactamase (see Current Opinion in Biotechnology 2005, 16:378–384
Figure 2c), generated in silico 7 1023 combinatorial mutants of these residues, and chose cut-offs to define the library of roughly 200 000 lowest-energy mutants that were then generated experimentally by mutagenesis and recombination. By selection against the antibiotic cefotaxime, the authors identified a mutant harbouring six mutations in the vicinity of the active site, with a 1280fold increase in resistance to cefotaxime that was attributed to an increase in relative substrate specificity toward cefotaxime versus ampicillin. The advantage of PDA is that it samples a vast sequence diversity and allows for multiple mutations to be identified simultaneously, which is particularly beneficial when the effect of multiple mutations is synergistic (non-additive) [20,44]. Furthermore, PDA generates mutations at the level of the amino acid sequence rather than at the level of the nucleotide sequence. Thus, there is no bias against mutations requiring two or three nucleotide modifications, contrary to the important bias that exists in standard random mutagenesis methods. This approach vastly increases explorable sequence space, but has not been specifically designed for improvement of enzyme activity. The second computational method for semi-rational and combinatorial design, which has been specifically applied to the design of enzyme active sites, was developed by Hellinga and co-workers [45]. This ground-breaking work, involving the prediction of mutations that are necessary for the introduction of catalytically active sites in non-catalytic protein scaffolds, allowed the authors to achieve the most dramatic success in computational enzyme design to date. In their report, Dwyer et al. [45] converted a non-catalytic ribose-binding protein (RBP) into an analog of triose phosphate isomerase (TIM). Their protein design algorithm first predicted mutations at the ribose-binding area to allow binding of the TIM substrate, dihydroxyacetone phosphate (DHAP) [46]. The algorithm then positioned a set of TIM catalytic residues in the new DHAP-binding pocket. Finally, 14 virtual constructs were experimentally generated and tested for TIM activity. Seven of the mutants possessed TIM activity greater than the background reaction. One of the seven designs was particularly active and through further improvements by computational design and random mutagenesis, allowed the generation of TIM analogs from a non-catalytic protein. The computational design allowed for the definition of 13 to 21 mutations, depending on the TIM analog, that introduce the necessary catalytic residues as well as residues forming a stereochemically complementary substrate-binding cavity. The most active TIM analogs displayed a 105–106-fold increase in reaction rate over background, which is the largest increase in reaction rate for a rationally designed enzyme to date [47–49], and was sufficiently active to support growth of TIM-deficient E. coli. An important contributing factor to the success of this www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 383
work results from their superior treatment of electrostatic interactions in a hetereogeneous protein environment [50]. Although this formidable achievement required prior knowledge of the precise chemical and steric requirements of the target activity, it opens the door to the broader creation of desirable catalytic activities within stable and well-behaved frameworks.
Conclusions Targeted, combinatorial semi-rational mutagenesis is proving highly effective for improving enzyme activities, as it readily allows the creation of neighbouring mutations, of multiple simultaneous mutations and of mutations requiring multiple nucleotide substitutions. This could be particularly advantageous in the modification of enzyme activities, as active-site mutations are frequently coupled and have synergistic effects [20,44]. Important developments in computational methodologies promise to vastly increase the searchable sequence space. Taken with a judicious choice of experimental input, semirational mutagenesis permits the experimenter to focus mutations in areas more likely to yield ‘lead’ results. These methodologies pave the way to exciting areas of enzyme research including efficient modification of existing activities, the development of new activities within existing frameworks, as well as the evolution of ‘promiscuous’ catalytic activities allowing their efficient exploitation [51,52].
Acknowledgements The authors thank Romas J Kazlauskas for his careful reading of the manuscript and his helpful comments.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest 1.
Radzicka A, Wolfenden R: A proficient enzyme. Science 1995, 267:90-93.
2.
Benkovic SJ, Hammes-Schiffer S: A perspective on enzyme catalysis. Science 2003, 301:1196-1202.
3.
Kirk O, Borchert TV, Fuglsang CC: Industrial enzyme applications. Curr Opin Biotechnol 2002, 13:345-351.
4.
Bornscheuer UT, Pohl M: Improved biocatalysts by directed evolution and rational protein design. Curr Opin Chem Biol 2001, 5:137-143.
5.
Scrutton NS, Berry A, Perham RN: Redesign of the coenzyme specificity of a dehydrogenase by protein engineering. Nature 1990, 343:38-43.
6.
Craik CS, Largman C, Fletcher T, Roczniak S, Barr PJ, Fletterick R, Rutter WJ: Redesigning trypsin: alteration of substrate specificity. Science 1985, 228:291-297.
9.
Cedrone F, Me´ nez A, Que´ me´ neur E: Tailoring new enzyme functions by rational redesign. Curr Opin Struct Biol 2000, 10:405-410.
10. Wang L, Jackson WC, Steinbach PA, Tsien RY: Evolution of new nonantibody proteins via iterative somatic hypermutation. Proc Natl Acad Sci USA 2004, 101:16745-16749. 11. Jaeger KE, Eggert T: Enantioselective biocatalysis optimized by directed evolution. Curr Opin Biotechnol 2004, 15:305-313. 12. Jestin JL, Kaminski PA: Directed enzyme evolution and selections for catalysis based on product formation. J Biotechnol 2004, 113:85-103. 13. Tao H, Cornish VW: Milestones in directed enzyme evolution. Curr Opin Chem Biol 2002, 6:858-864. 14. Williams GJ, Nelson AS, Berry A: Directed evolution of enzymes for biocatalysis and the life sciences. Cell Mol Life Sci 2004, 61:3034-3046. 15. Dalby PA: Optimising enzyme function by directed evolution. Curr Opin Struct Biol 2003, 13:500-505. 16. Chockalingam K, Chen Z, Katzenellenbogen JA, Zhao H: Directed evolution of specific receptor–ligand pairs for use in the creation of gene switches. Proc Natl Acad Sci USA 2005, 102:5691-5696. 17. Park S, Morley KL, Horsman GP, Holmquist M, Hult K, Kazlauskas RJ: Focusing mutations into the P. fluorescens esterase binding site increases enantioselectivity more effectively than distant mutations. Chem Biol 2005, 12:45-54. 18. Strausberg SL, Ruan B, Fisher KE, Alexander PA, Bryan PN: Directed coevolution of stability and catalytic activity in calcium-free subtilisin. Biochemistry 2005, 44:3272-3279. A great example of semi-rational design of enzymatic activity and stability based on structural and functional data previously obtained with subtilisin. Jointly rational and semi-rational, the approach is based on a stepwise increase in the activity of mutants obtained by a randomization of targeted residues near the active-site cavity. 19. Morley KL, Kazlauskas RJ: Improving enzyme properties: when are closer mutations better? Trends Biotechnol 2005, 23:231-237. 20. Mildvan AS: Inverse thinking about double mutants of enzymes. Biochemistry 2004, 43:14517-14520. 21. Santoro SW, Schultz PG: Directed evolution of the site specificity of Cre recombinase. Proc Natl Acad Sci USA 2002, 99:4185-4190. Using structural knowledge, residues of Cre recombinase from bacteriophage P1 that are in direct contact with variants of the loxP DNA substrate were subjected to a semi-rational saturation mutagenesis approach to successfully switch the substrate specificity of the Cre enzyme. A powerful screening method using FACS is also described. 22. Rui L, Cao L, Chen W, Reardon KF, Wood TK: Active site engineering of the epoxide hydrolase from Agrobacterium radiobacter AD1 to enhance aerobic mineralization of cis-1,2-dichloroethylene in cells expressing an evolved toluene ortho-monooxygenase. J Biol Chem 2004, 279:46810-46817. 23. Hill CM, Li WS, Thoden JB, Holden HM, Raushel FM: Enhanced degradation of chemical warfare agents through molecular engineering of the phosphotriesterase active site. J Am Chem Soc 2003, 125:8990-8991. 24. Wilming M, Iffland A, Tafelmeyer P, Arrivoli C, Saudan C, Johnsson K: Examining reactivity and specificity of cytochrome c peroxidase by using combinatorial mutagenesis. ChemBioChem 2002, 3:1097-1104.
7.
Carter P, Nilsson B, Burnier JP, Burdick D, Wells JA: Engineering subtilisin BPN’ for site-specific proteolysis. Proteins 1989, 6:240-248.
25. Antikainen NM, Hergenrother PJ, Harris MM, Corbett W, Martin SF: Altering substrate specificity of phosphatidylcholine-preferring phospholipase C of Bacillus cereus by random mutagenesis of the headgroup binding site. Biochemistry 2003, 42:1603-1610.
8.
Wells JA, Powers DB, Bott RR, Graycar TP, Estell DA: Designing substrate specificity by protein engineering of electrostatic interactions. Proc Natl Acad Sci USA 1987, 84:1219-1223.
26. Yew WS, Akana J, Wise EL, Rayment I, Gerlt JA: Evolution of enzymatic activities in the orotidine 5(-monophosphate decarboxylase suprafamily: enhancing the promiscuous D-arabino-hex-3-ulose 6-phosphate synthase reaction
www.sciencedirect.com
Current Opinion in Biotechnology 2005, 16:378–384
384 Protein technologies and commercial enzymes
catalyzed by 3-keto- L-gulonate 6-phosphate decarboxylase. Biochemistry 2005, 44:1807-1815. 27. Anderson JC, Wu N, Santoro SW, Lakshman V, King DS, Schultz PG: An expanded genetic code with a functional quadruplet codon. Proc Natl Acad Sci USA 2004, 101:7566-7571. 28. Schmitzer AR, Le´ pine F, Pelletier JN: Combinatorial exploration of the catalytic site of a drug-resistant dihydrofolate reductase: creating alternative functional configurations. Protein Eng Des Sel 2004, 17:809-819. 29. Cheon YH, Park HS, Kim JH, Kim Y, Kim HS: Manipulation of the active site loops of D-hydantoinase, a (b/a)8-barrel protein, for modulation of the substrate specificity. Biochemistry 2004, 43:7413-7420. 30. Peimbert M, Segovia L: Evolutionary engineering of a b-lactamase activity on a D-Ala D-Ala transpeptidase fold. Protein Eng 2003, 16:27-35. 31. Reetz MT: Controlling the enantioselectivity of enzymes by directed evolution: practical and theoretical ramifications. Proc Natl Acad Sci USA 2004, 101:5716-5722. An impressive example of enzyme evolution using a combination of semirational and random mutagenesis approaches to overcome the respective limitations of each method. The author consolidates the progression in modification of enantioselectivity in P. aeruginosa lipases from E = 1.1 (wild type) to E = 51, clearly describing improvements obtained with each mutational step. 32. Reetz MT, Wilensek S, Zha DX, Jaeger KE: Directed evolution of an enantioselective enzyme through combinatorial multiple-cassette mutagenesis. Angew Chem Int Ed Engl 2001, 40:3589-3591. 33. Koga Y, Kato K, Nakano H, Yamane T: Inverting enantioselectivity of Burkholderia cepacia KWI-56 lipase by combinatorial mutation and high-throughput screening using single-molecule PCR and in vitro expression. J Mol Biol 2003, 331:585-592. 34. Horsman GP, Liu AMF, Henke E, Bornscheuer UT, Kazlauskas RJ: Mutations in distant residues moderately increase the enantioselectivity of Pseudomonas fluorescens esterase towards methyl 3-bromo-2-methylpropanoate and ethyl 3-phenylbutyrate. Chemistry 2003, 9:1933-1939. 35. Lingen B, Grotzinger J, Kolter D, Kula MR, Pohl M: Improving the carboligase activity of benzoylformate decarboxylase from Pseudomonas putida by a combination of directed evolution and site-directed mutagenesis. Protein Eng 2002, 15:585-593. 36. Geddie ML, Matsumura I: Rapid evolution of b-glucuronidase specificity by saturation mutagenesis of an active site loop. J Biol Chem 2004, 279:26462-26468. 37. Sio CF, Riemens AM, van der Laan JM, Verhaert RM, Quax WJ: Directed evolution of a glutaryl acylase into an adipyl acylase. Eur J Biochem 2002, 269:4495-4504. 38. Shaner NC, Campbell RE, Steinbach PA, Giepmans BN, Palmer AE, Tsien RY: Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat Biotechnol 2004, 22:1567-1572.
Current Opinion in Biotechnology 2005, 16:378–384
Simultaneous saturation mutagenesis at many targeted positions coupled to epPCR allowed the authors to generate an impressive array of multiple fluorescent and stable monomer variants of Discosoma sp. fluorescent protein (DsRed) based solely on sequence comparison of mutants of interest. These stable monomers are now applicable to discrimination of cell types, transcriptional activities and/or fusion proteins. Although not oriented toward catalysis, this example nevertheless demonstrates the power of semi-rational approaches. 39. Lichtarge O, Bourne HR, Cohen FE: An evolutionary trace method defines binding surfaces common to protein families. J Mol Biol 1996, 257:342-358. 40. Lichtarge O, Yao H, Kristensen DM, Madabushi S, Mihalek I: Accurate and scalable identification of functional sites by evolutionary tracing. J Struct Funct Genomics 2003, 4:159-166. 41. Minshull J, Ness JE, Gustafsson C, Govindarajan S: Predicting enzyme function from protein sequence. Curr Opin Chem Biol 2005, 9:202-209. 42. Dahiyat BI: In silico design for protein stabilization. Curr Opin Biotechnol 1999, 10:387-390. 43. Hayes RJ, Bentzien J, Ary ML, Hwang MY, Jacinto JM, Vielmetter J, Kundu A, Dahiyat BI: Combining computational and experimental screening for rapid optimization of protein properties. Proc Natl Acad Sci USA 2002, 99:15926-15931. An innovative approach to the use of computational tools in protein engineering. The prescreen described here could be invaluable for the application of a semi-rational approach to enzyme improvement. 44. Voigt CA, Kauffman S, Wang ZG: Rational evolutionary design: the theory of in vitro protein evolution. Adv Protein Chem 2000, 55:79-160. 45. Dwyer MA, Looger LL, Hellinga HW: Computational design of a biologically active enzyme. Science 2004, 304:1967-1971. Computational design of an enzymatic activity in a protein scaffold of known structure. Ground-breaking in that it demonstrates the feasibility of creating new enzymatic activities where there was none by introducing mutations at or near the substrate-binding site. 46. Looger LL, Dwyer MA, Smith JJ, Hellinga HW: Computational design of receptor and sensor proteins with novel functions. Nature 2003, 423:185-190. 47. Hilvert D: Critical analysis of antibody catalysis. Annu Rev Biochem 2000, 69:751-793. 48. Bolon DN, Mayo SL: Enzyme-like proteins by computational design. Proc Natl Acad Sci USA 2001, 98:14274-14279. 49. Bolon DN, Voigt CA, Mayo SL: De novo design of biocatalysts. Curr Opin Chem Biol 2002, 6:125-129. 50. Wisz MS, Hellinga HW: An empirical model for electrostatic interactions in proteins incorporating multiple geometrydependent dielectric constants. Proteins 2003, 51:360-377. 51. Kazlauskas RJ: Enhancing catalytic promiscuity for biocatalysis. Curr Opin Chem Biol 2005, 9:195-201. 52. Aharoni A, Gaidukov L, Khersonsky O, Mc QGS, Roodveldt C, Tawfik DS: The ‘evolvability’ of promiscuous protein functions. Nat Genet 2005, 37:73-76.
www.sciencedirect.com
Current Opinion in Biotechnology 2005, 16:378–384 This review comes from a themed issue on Protein technologies and commercial enzymes Edited by Bernhard Hauer and Brian K Kay Available online 1st July 2005 0958-1669/$ – see front matter # 2005 Elsevier Ltd. All rights reserved. DOI 10.1016/j.copbio.2005.06.004
Introduction Enzymes are powerful catalysts that are able to increase reaction rates by up to 17 orders of magnitude [1,2]. Certain enzymes display perfect control over stereochemistry and regioselectivity, while others display a breadth of specificity; either feature may be attractive for industrial and synthetic applications [3]. However, few enzymes naturally catalyze the reactions that chemists need under conditions that are industrially convenient and economically advantageous. Many efforts are currently devoted to the modification of enzyme activities to meet the needs of today’s chemists. Multiple approaches have been developed to allow the identification of mutant enzymes possessing desirable Current Opinion in Biotechnology 2005, 16:378–384
qualities such as increased activity, modified specificity, selectivity or cofactor binding [4]. The earliest approach was rational design, which was used to modify the specificity of enzymes [5–9]. This approach requires an indepth knowledge of the structural features of the enzyme active site and their contribution to function. The complexity of the structure/function relationship in enzymes has proven to be the factor limiting the general application of rational design. More recently, directed evolution has proven to be a powerful tool for the modification of enzyme activities and has become the most widely used approach. Briefly, directed evolution consists of the low frequency introduction of randomly distributed mutations in a gene of interest, followed by selection of the mutated proteins possessing the desired properties. Recent advances in this area include the application of the immune system ‘hypermutation’ process, which is conducted entirely in live cells [10]. Directed evolution can enable the relatively rapid engineering of enzymes without requiring an in-depth understanding of structure/ function relationships, unlike rational design. The numerous methodologies and successes of directed evolution will not be detailed here as recent reviews have been devoted to the topic [11–15] (also see the article by AA Henry and FE Romesberg in this issue). Because large numbers of mutants must generally be screened to obtain a significant, desired effect on enzyme activity, the main limitation of directed evolution is the necessity of developing a high-throughput screening methodology that allows identification of the desired property under relevant conditions. Not all enzyme activities are readily amenable to developing a high-throughput screening method, nor are all screening methodologies easy to implement at the required scale. A further limitation of directed evolution relates to the random nature of the mutations introduced: in the case where functional information (from point mutations, random mutagenesis or deduction by sequence alignment) or structural information exists, it would be advantageous to exploit this by concentrating mutations where they might be the most effective [16]. Analysis of enzyme modification results indicates that the majority of mutations that beneficially affect certain enzyme properties (enantioselectivity, substrate specificity and new catalytic activities) are located in or near the active site, and more specifically near residues that are implicated in binding or catalysis [15,17,18,19]. Other properties (stability and activity) can be improved by mutations either near or far from the active site, although there are many more mutations far from the active site to test out [19]. www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 379
Table 1 Comparison of approaches for engineering enzyme activity. Rational design
Random mutagenesis
Semi-rational design
High-throughput screening or selection method Structural and/or functional information Sequence space exploration
Not essential Both essential Low
Essential Neither essential Moderate, random
Probability of obtaining synergistic mutations
Moderate
Low
Advantageous but not essential Either is sufficient Experimental: moderate, targeted Computational: vast, targeted High
Consequently, saturation mutagenesis (where all 20 natural amino acids are tested at residues in or near the active site) can be coupled to recombination of these mutations in order to increase the likelihood of beneficially modifying the catalytic activity relative to random mutagenesis approaches. This ‘semi-rational’ approach could be particularly advantageous in instances where no highthroughput screening method is available, based on the argument that a ‘smarter’ library can be built with the same number of mutants than with a random whole-gene mutagenesis approach [15] (Table 1; Figure 1). This review provides an overview of recent articles where the semi-rational design of enzyme activities
has been successfully demonstrated through the use of combinatorial mutagenesis biased toward the active site, based both on experimental and computational approaches.
Modifying enzyme activity using experimental semi-rational approaches The experimental combination of rational and random protein engineering approaches has been successfully applied toward the modification of enzyme activities. The following examples have been grouped according to the specific experimental approaches applied, which are guided by the amount and nature of structural and functional information available at the outset of the study.
Figure 1
Choice of method for enzyme engineering
Functional information may be gained from small library or point mutations
NO
NO
High-throughput screening available?
Structural/functional information available?
NO
YES
High-throughput screening available?
YES
YES WHOLE-GENE RANDOMIZATION
SEMI-RATIONAL DESIGN Rational design
SEMI-RATIONAL DESIGN Whole-gene randomization Current Opinion in Biotechnology
Selection of the preferred experimental approach for enzyme engineering based on the availability of experimental tools and prior knowledge of structure and function. Rational design, semi-rational design or whole-gene randomization each refer to multiple methodologies, as outlined in the text. The enzyme engineering approach that may have the greatest potential for success is in upper case letters, while alternative approaches are in lower case letters. www.sciencedirect.com
Current Opinion in Biotechnology 2005, 16:378–384
380 Protein technologies and commercial enzymes
Targeted randomization of defined residues based on structural knowledge
Structural information, when available, can allow one to target specific residues in direct contact with the substrate or near the active-site cavity for mutagenesis, alone or in combination. Although the choice of the positions to mutate remains rational in most cases, the choice of amino acids to be encoded can be broad; furthermore, the simultaneous randomization at targeted positions may result in synergistic effects [20] that could not have been predicted by mutating positions individually. Santoro and Schultz [21] have used targeted randomization to modify the substrate specificity of a Cre DNA recombinase from bacteriophage P1 that recognizes DNA sequences known as loxP sites. The researchers used structural information of the Cre–DNA complex to target two regions of the Cre recombinase in contact with loxP base pairs suspected to be important for recombination activity. By creating two distinct libraries (C1 and C2; see Figure 2a) of five and six simultaneously randomized residues in contact with variants of loxP sites (library size 108) and by subjecting the E. coli-transformed libraries to a powerful recombination assay of fluorescent reporter proteins sorted by FACS (fluorescence-activated cell sorting), they isolated a mutant that efficiently recombined a new loxP site not recognized by the wild-type enzyme. Moreover, this Cre mutant retained the ability to recognize the native loxP site. The authors favored this targeted library approach because they anticipated that a higher likelihood of success would be observed, while requiring fewer rounds of selective amplification and
mutagenesis compared with random mutagenesis of the whole gene. Similarly, in a successful attempt to investigate the detoxification role of epoxide hydrolases, Rui et al. [22] used a combination of rational design and saturation mutagenesis at targeted active-site positions to expand the substrate range of an epoxide hydrolase from Agrobacterium radiobacter AD1 (EchA) to include chlorinated epoxides. On the basis of a careful investigation of the active site by structural comparison to related enzymes, residues F108, C248, I219 and I111 (in single-letter amino acid code), in the vicinity of the catalytic triad, were separately randomized by saturation mutagenesis. As mutant C248I showed a slight increase in rate of cis1,2-dichloroethylene (cis-DCE) mineralization (2.7-fold increase with respect to the wild type), it was used as a template for successive rounds of saturation mutagenesis targeting residues F108 and I219. The approach generated the triple active-site area mutant F108L-I219LC248L, which displayed a 10-fold enhancement in cis-DCE mineralization relative to wild-type EchA. Several other groups [23–26] have also successfully conducted simultaneous or successive targeted randomization of multiple active-site positions based on structural knowledge, as an efficient means to circumvent limitations inherent either to site-directed mutagenesis or to whole-gene random mutagenesis. Recently, Schultz and colleagues [27] combinatorially randomized two activesite residues of an aminoacyl-tRNA synthetase as a step in the generation of an orthogonal synthetase/tRNA pair
Figure 2
Residues in or near the enzyme active site that were targeted for semi-rational combinatorial mutagenesis. The enzymes are shown as grey ribbon diagrams with space-filling in semi-transparent gray. (a) Cre DNA recombinase from bacteriophage P1 [21] (PDB accession number 1CRX). Residues from libraries C1 (six residues) and C2 (five residues) and the DNA substrate are colored yellow, green and blue, respectively. (b) PBP2X DD-transpeptidase from S. pneumonia [30] (PDB accession number 1QMF). The ten residues targeted for combinatorial site-directed mutagenesis are colored yellow and the three subsequent unforeseen beneficial mutations obtained by epPCR are colored green. Cefuroxime is acylated at the active-site residue Ser337 and is colored blue. (c) TEM-1 b-lactamase from E. coli [43] (PDB accession number 1FQG). The 19 active-site residues targeted for in silico mutagenesis using PDA are colored yellow. The benzylpenicillin antibiotic is acylated at the active-site residue Ser70 and is colored blue. Figure generated with Pymol Release 0.98 (http://www.pymol.org). Current Opinion in Biotechnology 2005, 16:378–384
www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 381
that efficiently and selectively incorporates an unnatural amino acid into proteins. In some cases, the structural information available is not sufficient to rationally select residues to be randomized. Thus, molecular modelling studies have been successfully used to identify residues most likely to be in contact with substrate molecules. Our group used this approach to perform combinatorial targeted mutagenesis of all 16 principal active-site residues of type II R67 dihydrofolate reductase, allowing the selection of new, highly modified active-site environments with activity at least as great as the native enzyme [28]. Similarly, the substrate specificity of Bacillus stearothermophilus SD1 D-hydantoinase was modified by a semi-rational mutagenesis approach targeting residues chosen on the basis of previous modelling studies [29]. Simultaneous random mutagenesis and site-saturation of defined residues
Based on evolutionary protein-fold similarity between two related enzyme families, Peimbert and Segovia [30] introduced a b-lactamase activity into PBP2X DDtranspeptidase from Streptococcus pneumonia through a combination of error-prone polymerase chain reaction (epPCR) and saturation mutagenesis at targeted positions. This approach was used to increase the odds of obtaining beneficial mutations at unpredictable locations in combination with mutations at carefully chosen residues. This approach should increase the return using the same number of mutants as a random whole-gene mutagenesis scheme. Thus, two active-site residues (F450 and W374) were targeted for saturation mutagenesis in combination with simultaneous amino acid replacements at eight positions near the active-site cavity (see Figure 2b). After the oligonucleotide-based mutagenesis, the gene was further amplified with a 0.8% random mutagenesis rate. This library was selected for cefotaxime resistance and mutants with 10-fold increased resistance relative to wild-type PBP2X were obtained. The random mutagenesis resulted in additional mutations at positions 312, 452 and 554, both proximal to and distal from the active site. To assess the impact of certain randomly inserted mutations, saturation mutagenesis was undertaken at positions 312, 336, 450 and 452. Although no further increase in cefotaxime resistance was observed compared with the original mutants, one new mutant showed b-lactamase activity without compromising the original DD-peptidase activity of PBP2X, thus generating a mutant with dual substrate specificity. Random mutagenesis followed by site-saturation of defined residues
The structural or functional information necessary to make rational choices for the residues to mutate is not always available. To circumvent this limitation, several research groups have efficiently undertaken rounds of www.sciencedirect.com
whole-gene randomization to provide ‘leads’ (i.e. residues identified as potentially advantageous when mutated) in conjunction with fine-tuning of the ‘lead mutants’ through site-saturation mutagenesis of the identified residues. Using such a cyclical random/targeted approach, Reetz and collaborators greatly improved the enantioselectivity (E) of Pseudomonas aeruginosa lipases toward a p-nitrophenyl ester in favour of the (2S) enantiomer (reviewed in [31]). Initial improvement from E = 1.1 (wild-type enzyme) to E = 11.3 was achieved after four rounds of epPCR. Using the mutational knowledge obtained by epPCR, they conducted saturation mutagenesis at given mutational ‘hot spots’ of the lipase, obtaining E = 20. Subsequent rounds of epPCR increased E to 25. In addition, they developed a modified version of Stemmer’s combinatorial multiple-cassette mutagenesis method [32] to mutate a defined region of the lipase (residues 160–163) as well as several ‘hot spots’ (residues 155 and 162). This semi-rational method yielded their most highly enantioselective lipase variant to date (E = 51). Other groups have recently demonstrated the power of similar semi-rational approaches toward the evolution of enantioselective enzymes. As a result of the multiple random mutagenesis steps, this approach has also generated many mutations far from the active-site cavity in addition to mutations designed within the active-site cavity [33–35]. Geddie and Matsumura [36] have modified the substrate specificity of Escherichia coli b-glucuronidase through combinatorial site-saturation mutagenesis of several ‘hot spots’, using knowledge previously obtained from DNA shuffling studies. The resulting mutants displayed up to a 70-fold increase in xylosidase activity and were further improved by several steps of whole-gene random mutagenesis using DNA shuffling, epPCR and Staggered Extension Process (StEP) to generate mutants displaying 100-fold improvement in xylosidase activity compared with their ancestor. Similar studies were undertaken to modify the enzymatic specificity and activity toward 7-aminodesacetoxycephalosporanic acid (7-ADCA) of a glutaryl acylase of Pseudomonas SY-77 [37] and to modify and improve fluorescent proteins from Discosoma sp. [38]. In all of these studies, whole-gene randomization was generally undertaken in a first round of mutagenesis, followed by ‘fine-tuning’ of the most interesting mutants through single or combinatorial site-saturation mutagenesis at targeted positions. Afterwards, mutants obtained were further improved using subsequent rounds of random mutagenesis. In the absence of structural or functional information, the ‘Evolutionary Trace’ method [39] can pinpoint residues to mutate. This method entails correlating evolutionary variations within a gene of interest with divergences in the phylogenetic tree of that sequence family. This has Current Opinion in Biotechnology 2005, 16:378–384
382 Protein technologies and commercial enzymes
been shown to reveal the relative functional importance of residues and to identify functional sites [40]. A recent review by Minshull et al. [41] provides a highly comprehensive overview of this and other approaches for identifying residues that may be functionally relevant. These recent examples demonstrate the power of structure-based semi-rational approaches targeting active-site residues in enzyme engineering by enhancing our capacity for rational design, while exploiting the advantageous aspects of random mutagenesis. We can thus cycle between light sampling of randomly distributed mutations and saturation mutagenesis at a limited number of positions likely to affect the property under study, increasing the likelihood of identifying beneficial, cooperative effects with respect to enzyme catalysis. Semi-rational and combinatorial design using computational approaches
Although semi-rational, combinatorial mutagenesis directed at active-site residues limits the number of variants generated relative to available sequence space and theoretically focuses on an area of sequence space where improvements have a greater likelihood of occurring, it can nonetheless generate greater numbers of mutants than can be screened. Indeed, combinatorial randomization of only five residues generates a library of 205 possibilities (3.2 106 mutants), too large a number for manual screening. Thus, to increase the power of semirational and combinatorial modification of enzyme activities, computational methods have been developed based on protein design algorithms. These methods can either perform a virtual screening of a vast library or can be applied to the design of enzyme active sites. The first method is the computational screening of mutant sequences of a virtual library. Computational screening can screen libraries of 1080 variants [42], allowing for a first layer of virtual screening to eliminate mutations inconsistent with the protein fold. This is an emerging area in protein design and has seldom yet been applied specifically to enzyme design. However, its resounding success and great promise merit discussion herein. Hayes et al. [43] developed a strategy for the computational screening of large libraries called Protein Design Automation (PDA), for prediction of the optimal sequence that can adopt a desired fold. PDA was used for prescreening large, virtual libraries of mutants (1023), thus decreasing the sequence space of interest by many orders of magnitude. PDA allows all, or a rationally defined set of residues, to change. The optimal sequence is chosen based on its lowest conformational energy and is used to identify other near-optimal sequences through Monte Carlo simulated annealing. The mutations that occur most frequently define the library to be experimentally screened. Using PDA, Hayes and colleagues pinpointed 19 residues of interest in TEM-1 b-lactamase (see Current Opinion in Biotechnology 2005, 16:378–384
Figure 2c), generated in silico 7 1023 combinatorial mutants of these residues, and chose cut-offs to define the library of roughly 200 000 lowest-energy mutants that were then generated experimentally by mutagenesis and recombination. By selection against the antibiotic cefotaxime, the authors identified a mutant harbouring six mutations in the vicinity of the active site, with a 1280fold increase in resistance to cefotaxime that was attributed to an increase in relative substrate specificity toward cefotaxime versus ampicillin. The advantage of PDA is that it samples a vast sequence diversity and allows for multiple mutations to be identified simultaneously, which is particularly beneficial when the effect of multiple mutations is synergistic (non-additive) [20,44]. Furthermore, PDA generates mutations at the level of the amino acid sequence rather than at the level of the nucleotide sequence. Thus, there is no bias against mutations requiring two or three nucleotide modifications, contrary to the important bias that exists in standard random mutagenesis methods. This approach vastly increases explorable sequence space, but has not been specifically designed for improvement of enzyme activity. The second computational method for semi-rational and combinatorial design, which has been specifically applied to the design of enzyme active sites, was developed by Hellinga and co-workers [45]. This ground-breaking work, involving the prediction of mutations that are necessary for the introduction of catalytically active sites in non-catalytic protein scaffolds, allowed the authors to achieve the most dramatic success in computational enzyme design to date. In their report, Dwyer et al. [45] converted a non-catalytic ribose-binding protein (RBP) into an analog of triose phosphate isomerase (TIM). Their protein design algorithm first predicted mutations at the ribose-binding area to allow binding of the TIM substrate, dihydroxyacetone phosphate (DHAP) [46]. The algorithm then positioned a set of TIM catalytic residues in the new DHAP-binding pocket. Finally, 14 virtual constructs were experimentally generated and tested for TIM activity. Seven of the mutants possessed TIM activity greater than the background reaction. One of the seven designs was particularly active and through further improvements by computational design and random mutagenesis, allowed the generation of TIM analogs from a non-catalytic protein. The computational design allowed for the definition of 13 to 21 mutations, depending on the TIM analog, that introduce the necessary catalytic residues as well as residues forming a stereochemically complementary substrate-binding cavity. The most active TIM analogs displayed a 105–106-fold increase in reaction rate over background, which is the largest increase in reaction rate for a rationally designed enzyme to date [47–49], and was sufficiently active to support growth of TIM-deficient E. coli. An important contributing factor to the success of this www.sciencedirect.com
Engineering enzyme activity Chica, Doucet and Pelletier 383
work results from their superior treatment of electrostatic interactions in a hetereogeneous protein environment [50]. Although this formidable achievement required prior knowledge of the precise chemical and steric requirements of the target activity, it opens the door to the broader creation of desirable catalytic activities within stable and well-behaved frameworks.
Conclusions Targeted, combinatorial semi-rational mutagenesis is proving highly effective for improving enzyme activities, as it readily allows the creation of neighbouring mutations, of multiple simultaneous mutations and of mutations requiring multiple nucleotide substitutions. This could be particularly advantageous in the modification of enzyme activities, as active-site mutations are frequently coupled and have synergistic effects [20,44]. Important developments in computational methodologies promise to vastly increase the searchable sequence space. Taken with a judicious choice of experimental input, semirational mutagenesis permits the experimenter to focus mutations in areas more likely to yield ‘lead’ results. These methodologies pave the way to exciting areas of enzyme research including efficient modification of existing activities, the development of new activities within existing frameworks, as well as the evolution of ‘promiscuous’ catalytic activities allowing their efficient exploitation [51,52].
Acknowledgements The authors thank Romas J Kazlauskas for his careful reading of the manuscript and his helpful comments.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest 1.
Radzicka A, Wolfenden R: A proficient enzyme. Science 1995, 267:90-93.
2.
Benkovic SJ, Hammes-Schiffer S: A perspective on enzyme catalysis. Science 2003, 301:1196-1202.
3.
Kirk O, Borchert TV, Fuglsang CC: Industrial enzyme applications. Curr Opin Biotechnol 2002, 13:345-351.
4.
Bornscheuer UT, Pohl M: Improved biocatalysts by directed evolution and rational protein design. Curr Opin Chem Biol 2001, 5:137-143.
5.
Scrutton NS, Berry A, Perham RN: Redesign of the coenzyme specificity of a dehydrogenase by protein engineering. Nature 1990, 343:38-43.
6.
Craik CS, Largman C, Fletcher T, Roczniak S, Barr PJ, Fletterick R, Rutter WJ: Redesigning trypsin: alteration of substrate specificity. Science 1985, 228:291-297.
9.
Cedrone F, Me´ nez A, Que´ me´ neur E: Tailoring new enzyme functions by rational redesign. Curr Opin Struct Biol 2000, 10:405-410.
10. Wang L, Jackson WC, Steinbach PA, Tsien RY: Evolution of new nonantibody proteins via iterative somatic hypermutation. Proc Natl Acad Sci USA 2004, 101:16745-16749. 11. Jaeger KE, Eggert T: Enantioselective biocatalysis optimized by directed evolution. Curr Opin Biotechnol 2004, 15:305-313. 12. Jestin JL, Kaminski PA: Directed enzyme evolution and selections for catalysis based on product formation. J Biotechnol 2004, 113:85-103. 13. Tao H, Cornish VW: Milestones in directed enzyme evolution. Curr Opin Chem Biol 2002, 6:858-864. 14. Williams GJ, Nelson AS, Berry A: Directed evolution of enzymes for biocatalysis and the life sciences. Cell Mol Life Sci 2004, 61:3034-3046. 15. Dalby PA: Optimising enzyme function by directed evolution. Curr Opin Struct Biol 2003, 13:500-505. 16. Chockalingam K, Chen Z, Katzenellenbogen JA, Zhao H: Directed evolution of specific receptor–ligand pairs for use in the creation of gene switches. Proc Natl Acad Sci USA 2005, 102:5691-5696. 17. Park S, Morley KL, Horsman GP, Holmquist M, Hult K, Kazlauskas RJ: Focusing mutations into the P. fluorescens esterase binding site increases enantioselectivity more effectively than distant mutations. Chem Biol 2005, 12:45-54. 18. Strausberg SL, Ruan B, Fisher KE, Alexander PA, Bryan PN: Directed coevolution of stability and catalytic activity in calcium-free subtilisin. Biochemistry 2005, 44:3272-3279. A great example of semi-rational design of enzymatic activity and stability based on structural and functional data previously obtained with subtilisin. Jointly rational and semi-rational, the approach is based on a stepwise increase in the activity of mutants obtained by a randomization of targeted residues near the active-site cavity. 19. Morley KL, Kazlauskas RJ: Improving enzyme properties: when are closer mutations better? Trends Biotechnol 2005, 23:231-237. 20. Mildvan AS: Inverse thinking about double mutants of enzymes. Biochemistry 2004, 43:14517-14520. 21. Santoro SW, Schultz PG: Directed evolution of the site specificity of Cre recombinase. Proc Natl Acad Sci USA 2002, 99:4185-4190. Using structural knowledge, residues of Cre recombinase from bacteriophage P1 that are in direct contact with variants of the loxP DNA substrate were subjected to a semi-rational saturation mutagenesis approach to successfully switch the substrate specificity of the Cre enzyme. A powerful screening method using FACS is also described. 22. Rui L, Cao L, Chen W, Reardon KF, Wood TK: Active site engineering of the epoxide hydrolase from Agrobacterium radiobacter AD1 to enhance aerobic mineralization of cis-1,2-dichloroethylene in cells expressing an evolved toluene ortho-monooxygenase. J Biol Chem 2004, 279:46810-46817. 23. Hill CM, Li WS, Thoden JB, Holden HM, Raushel FM: Enhanced degradation of chemical warfare agents through molecular engineering of the phosphotriesterase active site. J Am Chem Soc 2003, 125:8990-8991. 24. Wilming M, Iffland A, Tafelmeyer P, Arrivoli C, Saudan C, Johnsson K: Examining reactivity and specificity of cytochrome c peroxidase by using combinatorial mutagenesis. ChemBioChem 2002, 3:1097-1104.
7.
Carter P, Nilsson B, Burnier JP, Burdick D, Wells JA: Engineering subtilisin BPN’ for site-specific proteolysis. Proteins 1989, 6:240-248.
25. Antikainen NM, Hergenrother PJ, Harris MM, Corbett W, Martin SF: Altering substrate specificity of phosphatidylcholine-preferring phospholipase C of Bacillus cereus by random mutagenesis of the headgroup binding site. Biochemistry 2003, 42:1603-1610.
8.
Wells JA, Powers DB, Bott RR, Graycar TP, Estell DA: Designing substrate specificity by protein engineering of electrostatic interactions. Proc Natl Acad Sci USA 1987, 84:1219-1223.
26. Yew WS, Akana J, Wise EL, Rayment I, Gerlt JA: Evolution of enzymatic activities in the orotidine 5(-monophosphate decarboxylase suprafamily: enhancing the promiscuous D-arabino-hex-3-ulose 6-phosphate synthase reaction
www.sciencedirect.com
Current Opinion in Biotechnology 2005, 16:378–384
384 Protein technologies and commercial enzymes
catalyzed by 3-keto- L-gulonate 6-phosphate decarboxylase. Biochemistry 2005, 44:1807-1815. 27. Anderson JC, Wu N, Santoro SW, Lakshman V, King DS, Schultz PG: An expanded genetic code with a functional quadruplet codon. Proc Natl Acad Sci USA 2004, 101:7566-7571. 28. Schmitzer AR, Le´ pine F, Pelletier JN: Combinatorial exploration of the catalytic site of a drug-resistant dihydrofolate reductase: creating alternative functional configurations. Protein Eng Des Sel 2004, 17:809-819. 29. Cheon YH, Park HS, Kim JH, Kim Y, Kim HS: Manipulation of the active site loops of D-hydantoinase, a (b/a)8-barrel protein, for modulation of the substrate specificity. Biochemistry 2004, 43:7413-7420. 30. Peimbert M, Segovia L: Evolutionary engineering of a b-lactamase activity on a D-Ala D-Ala transpeptidase fold. Protein Eng 2003, 16:27-35. 31. Reetz MT: Controlling the enantioselectivity of enzymes by directed evolution: practical and theoretical ramifications. Proc Natl Acad Sci USA 2004, 101:5716-5722. An impressive example of enzyme evolution using a combination of semirational and random mutagenesis approaches to overcome the respective limitations of each method. The author consolidates the progression in modification of enantioselectivity in P. aeruginosa lipases from E = 1.1 (wild type) to E = 51, clearly describing improvements obtained with each mutational step. 32. Reetz MT, Wilensek S, Zha DX, Jaeger KE: Directed evolution of an enantioselective enzyme through combinatorial multiple-cassette mutagenesis. Angew Chem Int Ed Engl 2001, 40:3589-3591. 33. Koga Y, Kato K, Nakano H, Yamane T: Inverting enantioselectivity of Burkholderia cepacia KWI-56 lipase by combinatorial mutation and high-throughput screening using single-molecule PCR and in vitro expression. J Mol Biol 2003, 331:585-592. 34. Horsman GP, Liu AMF, Henke E, Bornscheuer UT, Kazlauskas RJ: Mutations in distant residues moderately increase the enantioselectivity of Pseudomonas fluorescens esterase towards methyl 3-bromo-2-methylpropanoate and ethyl 3-phenylbutyrate. Chemistry 2003, 9:1933-1939. 35. Lingen B, Grotzinger J, Kolter D, Kula MR, Pohl M: Improving the carboligase activity of benzoylformate decarboxylase from Pseudomonas putida by a combination of directed evolution and site-directed mutagenesis. Protein Eng 2002, 15:585-593. 36. Geddie ML, Matsumura I: Rapid evolution of b-glucuronidase specificity by saturation mutagenesis of an active site loop. J Biol Chem 2004, 279:26462-26468. 37. Sio CF, Riemens AM, van der Laan JM, Verhaert RM, Quax WJ: Directed evolution of a glutaryl acylase into an adipyl acylase. Eur J Biochem 2002, 269:4495-4504. 38. Shaner NC, Campbell RE, Steinbach PA, Giepmans BN, Palmer AE, Tsien RY: Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat Biotechnol 2004, 22:1567-1572.
Current Opinion in Biotechnology 2005, 16:378–384
Simultaneous saturation mutagenesis at many targeted positions coupled to epPCR allowed the authors to generate an impressive array of multiple fluorescent and stable monomer variants of Discosoma sp. fluorescent protein (DsRed) based solely on sequence comparison of mutants of interest. These stable monomers are now applicable to discrimination of cell types, transcriptional activities and/or fusion proteins. Although not oriented toward catalysis, this example nevertheless demonstrates the power of semi-rational approaches. 39. Lichtarge O, Bourne HR, Cohen FE: An evolutionary trace method defines binding surfaces common to protein families. J Mol Biol 1996, 257:342-358. 40. Lichtarge O, Yao H, Kristensen DM, Madabushi S, Mihalek I: Accurate and scalable identification of functional sites by evolutionary tracing. J Struct Funct Genomics 2003, 4:159-166. 41. Minshull J, Ness JE, Gustafsson C, Govindarajan S: Predicting enzyme function from protein sequence. Curr Opin Chem Biol 2005, 9:202-209. 42. Dahiyat BI: In silico design for protein stabilization. Curr Opin Biotechnol 1999, 10:387-390. 43. Hayes RJ, Bentzien J, Ary ML, Hwang MY, Jacinto JM, Vielmetter J, Kundu A, Dahiyat BI: Combining computational and experimental screening for rapid optimization of protein properties. Proc Natl Acad Sci USA 2002, 99:15926-15931. An innovative approach to the use of computational tools in protein engineering. The prescreen described here could be invaluable for the application of a semi-rational approach to enzyme improvement. 44. Voigt CA, Kauffman S, Wang ZG: Rational evolutionary design: the theory of in vitro protein evolution. Adv Protein Chem 2000, 55:79-160. 45. Dwyer MA, Looger LL, Hellinga HW: Computational design of a biologically active enzyme. Science 2004, 304:1967-1971. Computational design of an enzymatic activity in a protein scaffold of known structure. Ground-breaking in that it demonstrates the feasibility of creating new enzymatic activities where there was none by introducing mutations at or near the substrate-binding site. 46. Looger LL, Dwyer MA, Smith JJ, Hellinga HW: Computational design of receptor and sensor proteins with novel functions. Nature 2003, 423:185-190. 47. Hilvert D: Critical analysis of antibody catalysis. Annu Rev Biochem 2000, 69:751-793. 48. Bolon DN, Mayo SL: Enzyme-like proteins by computational design. Proc Natl Acad Sci USA 2001, 98:14274-14279. 49. Bolon DN, Voigt CA, Mayo SL: De novo design of biocatalysts. Curr Opin Chem Biol 2002, 6:125-129. 50. Wisz MS, Hellinga HW: An empirical model for electrostatic interactions in proteins incorporating multiple geometrydependent dielectric constants. Proteins 2003, 51:360-377. 51. Kazlauskas RJ: Enhancing catalytic promiscuity for biocatalysis. Curr Opin Chem Biol 2005, 9:195-201. 52. Aharoni A, Gaidukov L, Khersonsky O, Mc QGS, Roodveldt C, Tawfik DS: The ‘evolvability’ of promiscuous protein functions. Nat Genet 2005, 37:73-76.
www.sciencedirect.com